LTT Labs Article - What's up with UPSs? Testing UPS Output
Posted by LabsLucas@reddit | hardware | View on Reddit | 28 comments
Our company has always had many UPSs around for the convenience and business case of not suddenly losing a ton of work. We've been intrigued to check them out further, but we've been wary of connecting any of them to measurement equipment considering the high voltages involved. There is a serious potential they could damage equipment or ourselves.
Despite all that, we're throwing caution to the wind to check out some UPSs from around the office. There are so many directions that UPS/surge testing could go so this article will cover the test setup and interesting exploration results.
Wait_for_BM@reddit
Does the waveform really matter that much when everything inside uses DC.
PC PSU rectify AC and have power factor correction to internal DC bus (bulk storage cap) then convert it to lower voltage. They really should look at the DC bus inside a PSU and see if there are droops/noise/glitches during the transfer.
The AC zero crossing really matter for electrical noise that doesn't affect a PC If the power already back out everywhere, who cares?
RainyInSAndreas@reddit
I had to buy a pure sine wave UPS after hearing loud buzzing from MSI AI1300P with the approximate sine wave UPS.
APC sell a gaming line that advertises sine wave.
https://www.se.com/us/en/residential/apc-home-and-small-business/gaming-ups-xbox-ps4-console-pc/sine-wave-for-gaming/
Innocent-bystandr@reddit
"gaming" UPS. What a world we live in.
siazdghw@reddit
I can't blame them for trying to tap into a new market.
While it does nothing special, I bet like 99.999% of gamers don't have a UPS and the number of gamers who are even aware they could have a UPS is probably similar.
Also the UPS market probably isn't growing very much? Between laptops replacing desktops and autosaves and cloudsaves becoming normal for most apps and games, having an extra 10 mins of runtime isn't that useful anymore for most consumers.
jones_supa@reddit
There is also the simple fact that the power grids in many locations are much more robust these days. If there are practically no power outages, there is not much need for an UPS.
RainyInSAndreas@reddit
They do have the rare combination of line-interactive with pure sine wave output.
I had to make do with a serve-grade UPS for the same. SUA1500 which is much more bulkier and keeps the fan always on, albeit at power volume than the online UPS.
-Gh0st96-@reddit
It's actually advertised "For gamers" to be fair
VforVictorian@reddit
I'm not sure how much it is still an issue, but I know in the past I've heard some PSUs with active power factor correction not working correctly on UPSs without a clean sine wave output.
RealThanny@reddit
PC's aren't the only thing plugged into a UPS. Waveform does matter for some things, such as anything with an AC motor in it.
S_A_N_D_@reddit
My understanding is that ac motors are inductive loads which shouldn't be plugged into normal UPS'. You need special ones designed to handle the inrush current which likely perform better than normal consumer grade versions.
Wait_for_BM@reddit
but we are on hardware subs with a lot of focus on computer hardware. The UPS they are testing are mostly for backup for short term computer use as that's what the reviewer has in their lab.
People are unlikely hooking them up to fridges. A fully pack fridge could handle a few hours of outages on insulation. Some of the traditional AC motors are now going to Brushless DC motor e.g. fridge compressors.
TheGreenTormentor@reddit
While it’s true that pure sine is really only necessary for going through iron, the HF on a square still isn’t great for switching supplies. It’s a bit harder on all your filtering components and it’s possible some of that noise will still get through. It’s not a massive deal by any means but who knows, could shorten their life a little bit.
IMO if your UPS really is just for quick downtime then who gives a shit, but if you expect to run stuff on it for hours then get a pure sine unit. Being able to plug a fan into it and have it not burn up is a nice bonus. If you really want to skip the middleman then put high voltage DC straight into the IEC socket (or don’t).
Wait_for_BM@reddit
The type of UPS they have in the review aren't going to give 8-10 hours run time.
jimmy5853@reddit
The waveform stuff is interesting but for most PC users, stepped approximation is fine for short outages. Pure sine matters more for motors or sensitive audio gear. Good to see someone actually measuring instead of just repeating marketing specs though. Nice work.
zboarderz@reddit
Pure sine wave also matters for sensitive network gear or NAS devices. For example, from my understanding, ubiquitis NAS’s won’t work with their own UPS because it has a simulated sine wave and not a full sine wave.
VenditatioDelendaEst@reddit
I wouldn't call that "sensitive". I would call it, "unwillingness to pay for competent power supply engineering".
There is nothing specific to NAS or network gear that leads to that outcome.
VenditatioDelendaEst@reddit
May I suggest something?
Calculate and report the missing V^2 * seconds, which has units of joule-ohm. Maybe 99%-ile worst case. This gives you a measure of how much energy the connected equipment's power supply bus caps have to buffer to ride through the switching event. Compared to plain milliseconds, that measure is independent of where in the cycle the mains stops and where the UPS starts.
You would set your scope to square and integrate the output voltage signal. Then you'd subtract the value some an number of (ideal) periods, 166.7 ms after switchover from the value before. Then you subtract that from 2400 JΩ, which is what 10 cycles of 120V 60Hz would give you.
"But my connected equipment's power supply isn't a resistor?!" -- power factor correction works to make it act like one at 60 Hz.
WorldClassPianist@reddit
Can you test the LiFePO4 ups like the goldenmate ones?
Freaky_Freddy@reddit
Could be interesting to test some of the Anker SOLIX products (C300, C1000)
They're batteries but they claim they can be used as UPSs
tehbilly@reddit
I've not found much to substantiate their claims, but if I could get a good UPS that wasn't constantly fed by those tiny ass lead batteries I'd buy it in a heartbeat.
I hate the current lineup of desktop/workstation grade solutions but I absolutely can't rawdog my utility company's brownout-or-mini-blackout-roulette service.
sidesslidingslowly@reddit
The newest Anker solix and Eco flow models have 10ms crossover and work perfectly.
I use a ecoflow river 3 plus myself for my home lab (server + Synology+unifi setup) that pulls 200w constant, and it's always worked perfectly when I pull the plug on the eco-flow to cycle the battery level.
Fadelesstriker@reddit
Glad this is being covered!
I am staying in South Africa where power outtages are very prevalent.
I’ve had my previous Evga PSU fry during a surge during a scheduled power outtage (loadhsedding).
Which also killed a wifi networkcard plugged in via PCI-E.
Looking to build a high end system soon but am refusing to do so unless I can guarantee clean power. My current apartment has regular brown-outs and voltage drops…
I am considering getting an online ups, but they are bulky and loud, though very easy to come by on the grey market.
I would be interested to see how a pure-sinewave online ups would compare in these tests.
Slicethatbread@reddit
This is pretty cool, is there a chance that it could expand to other brands? The only brands I knew of the top of my head were APC and CyberPower, just asking because I'm wondering if APC is known to be better or if you had them because you were able to get better deals on them? Thanks for posting!
Creative-Macaroon324@reddit
I’ve been using the Bluetti Elite 200 V2 as my UPS, and the 15ms switchover is basically unnoticeable in real use, and enough 2kWh headroom to comfortably handles my desktop and monitors setup perfectly. It saved me several times during recent outages.
BuchMaister@reddit
Nice testing Lucas I enjoyed reading it, so next time you will explore THD and FD view ?
LabsLucas@reddit (OP)
Thank you! I do plan to look at the THD and FD for at least part/some of this, probably once I get some better/safer probing purchased and I can look at the current as well.
As I said in the article, there are so many different ways I can take this testing so I'm still deciding on the most informative path forward.
Numerlor@reddit
This is nice to have for some idea on measured differences between the units. I was recently looking to up my ups as I'm getting very close to maxing it out, and the range of prices for a given power rating is huge.
For 1200W I'm seeing units that star at 140€ whhich is even less than the 1000w eaton ellipse pro I have right now cost me, while something that looks more comparable in quality to the eaton is over twice the price at 300-350€
Netblock@reddit
That unloaded peak looks an awful lot like a voltage transition without droop/AVP. Or perhaps something regarding inductance smooths it out?
(For rotation-based AC generation, the output frequency changes based on the load; I don't think you'll see that with inverter-based generation)